LSTM-driven deep learning approach for accurate brain tumor detection in MRI data

Abstract

Brain tumor detection is a critical step in neuro-oncology, significantly influencing diagnosis, treatment planning, and patient prognosis. Magnetic Resonance Imaging (MRI) remains the preferred imaging modality for identifying brain tumors due to its high spatial resolution and non-invasive nature. However, manual interpretation of MRI scans is time-consuming and prone to variability. This study proposes an advanced deep learning framework based on Long Short-Term Memory (LSTM) networks to enhance the accuracy of brain tumor detection from MRI data. The model leverages LSTM’s ability to capture spatial and sequential dependencies across MRI slices, allowing effective differentiation among tumor types such as meningioma, glioma, and pituitary tumors. Using the BraTS 2018 dataset and additional MRI images from Figshare, the algorithm was trained and validated on over 8,000 annotated MRI slices. Experimental results demonstrate that the proposed approach achieves superior performance compared to conventional machine learning and convolutional neural network models, exhibiting high sensitivity, specificity, and overall accuracy. The model also shows resilience against noise and image variability, critical for practical clinical applications. This research highlights the potential of LSTM-based architectures in automating and improving brain tumor diagnosis, reducing radiologist workload, and enabling timely clinical decisions. Future work will focus on incorporating multimodal imaging and expanding dataset diversity to further enhance the system’s generalizability and robustness.